1206 MIIC and low grade serous ovarian cancer

Introduction/Background

Low grade serous ovarian cancer (LGSOC) is a rare entity which risk factors, treatment response and risk of relapse are partially understood. Currently available data suffer from small sample size and heterogeneous management, leading to limited knowledge. Mutlivariation Information-based Inductive Causation (MIIC) is a network learning method, able to analyze and exploit simultaneously and exhaustively a large number of patients data, to identify non visible correlation, without an a priori classification on the type of reconstructed network (causal or non-causal). The aim of this study was to identify new and unknown association on patients with a LGSOC using the MIIC algorithm, in order to develop new hypothesis.

Methodology

We conducted a multicenter retrospective study in 31 French healthcare centers between 2010 and 2017. We included all patients with a LGSOC, and collected clinical, histological, molecular, surgical and treatment data. The MIIC algorithm was applied to this database.

Results

317 patients were included. Variables were selected and pre-processed as follow: Center, age, menopausal status, tabacco consumption, abdominal surgery history, BMI, previous lesion (borderline or de novo), ascitis, initialCA125 level, initial Peritoneal Carcinosis Index, sus mesocolic involvment, digestive resection, wide peritonectomy, number of nodes removed, positive nodes, Complete resection,FIGO stage, Estrogen Receptor, Progesteron Receptor, somatic BRCA mutation, BRAF mutation, KRAS mutation, MicroSatellite Instability, chemotherapy, Hyperthermic IntraPeritoneal Chemotherapy, Bevacizumab, endocrine therapy, recurrence and death status. Known or obvious associations such as age and menopausal status, ascitis, CA125, allowed us to perform quality control of the algorithm. In addition, new associations that were previously little known or unknown, and still unexplored, have been highlighted, such as relationship between menopausal status and wide peritonectomy or nodal involvement.

Conclusion

The use of MIIC algorithm on a large LGSOC database has enabled the identification of interesting hypothesis, and future research topics.

Disclosures

The authors have no conflict of interest to declare.